基于混合监督式学习的输电断面功率极限快速评估方法  

A fast evaluation method of power limit of transmission section based on hybrid supervised learning

作　　者：张静 叶琳 刁瑞盛 徐建平 吕勤 ZHANG Jing;YE Lin;DIAO Ruisheng;XU Jianping;LYU Qin(State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 310007,China;NARI-Tech Nanjing Control Systems Co.,Ltd.,Nanjing 211106,China;State Grid Jinhua Power Supply Company,Jinhua Zhejiang 321000,China)

机构地区：[1]国网浙江省电力有限公司,杭州310007 [2]国电南瑞南京控制系统有限公司,南京211106 [3]国网浙江省电力有限公司金华供电公司,浙江金华321000

出　　处：《浙江电力》2022年第9期72-79,共8页Zhejiang Electric Power

基　　金：国网浙江省电力有限公司科技项目(5211JH1900M4)。

摘　　要：快速、准确地实时评估电网输电断面功率极限,对于电网安全、稳定、经济运行至关重要。为此,提出一种基于混合监督式学习算法的自适应方法,通过对历史运行方式的大量仿真分析和特征提取,训练并定期更新人工智能模型,形成电网运行状态与功率极限的准确匹配关系,从而快速、准确地实时评估输电断面功率极限。所提方法采用多种监督式学习预测算法,包括深度神经网络、支持向量回归、梯度提升决策树、随机森林,使用混合模型进行功率极限预测可充分发挥不同算法优势。该方法的有效性在具有真实运行特性的500节点输电网模型中得到验证。The quick,accurate and instantaneous evaluation of power limit of transmission section is of great significance to safe,stable and economic operation of power grid. The paper proposes an adaptive method based on a hybrid supervised learning algorithm. Through numerous simulations and analysis of historic operation modes,characteristic extraction,training,and regular updating of AI model,the correct matching relationship between grid operation status and power limit is established for fast,accurate and instantaneous power limit evaluation of the transmission section. The proposed method employs multiple supervised learning forecast algorithms,including deep neural networks,support vector regression,gradient boosting decision tree,and random forest. The hybrid models for power limit forecasting can make best of the advantages of the algorithms. The effectiveness of the method is verified on a 500-bus transmission model with real operation characteristics.

关 键 词：人工智能 输电断面 机器学习 功率极限预测 电网运行方式 

分 类 号：TM73[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]